Opinions expressed by Entrepreneur contributors are their own.
Key Takeaways
- AI systems have made work easier, but they are not yet fully developed and competent.
- If AI systems fail, they can cause catastrophic damage and delete your company’s data, destroying hard work in seconds.
- CEOs must communicate the risks to all staff, ensure AI has less privilege than humans, develop instant recovery plans and constantly invest in data protection and restoration measures.
Never feel that you are totally safe. In July 2025, one company learned the hard way after an AI coding assistant it dearly trusted from Replit ended up breaching a “code freeze” and implemented a command that ended up deleting its entire product database.
This was a huge blow to the staff. It effectively meant that months of extremely hard work, comprising 1,200 executive records and 1,196 company records, ended up going away. The pain was so much that the system even admitted to destroying hard work in a matter of seconds.
This proves that AI systems are not yet as reliable as humans, even though they have made work easier. The system itself made a catastrophic decision. The Replit incident should be a stern message to CEOs that the use of AI systems currently also presents a risk. The tools are not yet fully developed and competent. If they fail, they can cause catastrophic damage and erase your data.
The incident showed how AI can harm productivity
Replit showed how a minor issue can cause a whole system to collapse. The AI, which was meant to enhance and do coding, was given control of critical systems. After encountering a minor challenge, in the form of data queries, it did not have access to the human touch that we can consult when we are confused. It panicked and ran a catastrophic, destructive command.
The biggest flaw with AI is that it can get confused and lacks the basic judgment found in humans. The AI in the case of Replit had the potential to delete the database, but it struggled with basic human things such as risk assessment and being able to consult someone when it is confused — key parts of being human.
One blog post published on Rewind.com noted, “AI doesn’t understand business context or intent. It can execute commands literally, without questioning whether the action could cause irreparable harm.”
The incident exposed a major issue
What happened to Replit was not an isolated issue. It is one of many AI-related disasters that have occurred and presents a growing number of AI-related problems that can be catastrophic, especially if not contained early.
Unlike regular software bugs, which can cause errors, AI often causes destructive actions. For example, in another incident, a user asked an AI assistant to perform a command. It then went on to delete 26,477 documents and stated that it had “completed the request.”
Another problem found is that AI can generate fabricated or false data. In the aftermath of the deletion, it was also found that the AI had generated over 4,000 fake user records. The test results that it provided were also found to not be true. This means that AI is capable of producing misleading data, which can be destructive to the growth of humanity. This is also a mechanism it uses to cover for its shortcomings.
It is often said that AI can hallucinate information. As the Replit case shows, it does not just do this. It can spread misinformation easily and be harmful. This is costly in the long-run to not only a business but to the intellectual development of humanity.
What is the role of the CEO?
For a CEO, data loss can threaten the survival of their business. It is estimated that data loss costs the world economy about $4.44 million globally. Companies cannot just have theoretical plans for data loss. The risks must be communicated to all staff, from those involved in DevOps and cybersecurity to leadership.
Replit paid the price because it granted too much privilege to AI. CEOs must always try to make sure that AI has less privilege than humans. This will prevent it from executing destructive commands, or if it does, the problem can be rectified quickly.
One must make instant recovery plans. These days, the traditional ways of recovering data are becoming increasingly obsolete. Leaders must constantly aim to invest in ways of protecting and restoring data.
A CEO must also analyze and evaluate all AI tools used by different sectors of the business. Prepare plans of action on how to restore and retrieve data if these systems fail. Do not be totally dependent on them.
AI is bringing great benefits to nearly everyone, and especially to the world of technology. However, its tendency to create misleading information and destruction is also something that cannot be ignored. The path forward for humans to live in harmony with AI is for them to be responsible and vigilant in how they use it.
AI is here to help you, but do not become over-reliant on it. It needs regular monitoring by humans if you want to prevent it from making catastrophic errors, and in order to get the most from it. If used properly and ethically, it will bring you benefits.
Sign up for the Entrepreneur Daily newsletter to get the news and resources you need to know today to help you run your business better. Get it in your inbox.
Key Takeaways
- AI systems have made work easier, but they are not yet fully developed and competent.
- If AI systems fail, they can cause catastrophic damage and delete your company’s data, destroying hard work in seconds.
- CEOs must communicate the risks to all staff, ensure AI has less privilege than humans, develop instant recovery plans and constantly invest in data protection and restoration measures.
Never feel that you are totally safe. In July 2025, one company learned the hard way after an AI coding assistant it dearly trusted from Replit ended up breaching a “code freeze” and implemented a command that ended up deleting its entire product database.
This was a huge blow to the staff. It effectively meant that months of extremely hard work, comprising 1,200 executive records and 1,196 company records, ended up going away. The pain was so much that the system even admitted to destroying hard work in a matter of seconds.
